TrajAD: Trajectory Anomaly Detection for Trustworthy LLM Agents
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914310659244032 |
|---|---|
| author | Liu, Yibing Zhang, Chong Han, Zhongyi Liu, Hansong Wang, Yong Yu, Yang Wang, Xiaoyan Yin, Yilong |
| author_facet | Liu, Yibing Zhang, Chong Han, Zhongyi Liu, Hansong Wang, Yong Yu, Yang Wang, Xiaoyan Yin, Yilong |
| contents | We address the problem of runtime trajectory anomaly detection, a critical capability for enabling trustworthy LLM agents. Current safety measures predominantly focus on static input/output filtering. However, we argue that ensuring LLM agents reliability requires auditing the intermediate execution process. In this work, we formulate the task of Trajectory Anomaly Detection. The goal is not merely detection, but precise error localization. This capability is essential for enabling efficient rollback-and-retry. To achieve this, we construct TrajBench, a dataset synthesized via a perturb-and-complete strategy to cover diverse procedural anomalies. Using this benchmark, we investigate the capability of models in process supervision. We observe that general-purpose LLMs, even with zero-shot prompting, struggle to identify and localize these anomalies. This reveals that generalized capabilities do not automatically translate to process reliability. To address this, we propose TrajAD, a specialized verifier trained with fine-grained process supervision. Our approach outperforms baselines, demonstrating that specialized supervision is essential for building trustworthy agents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_06443 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TrajAD: Trajectory Anomaly Detection for Trustworthy LLM Agents Liu, Yibing Zhang, Chong Han, Zhongyi Liu, Hansong Wang, Yong Yu, Yang Wang, Xiaoyan Yin, Yilong Cryptography and Security Artificial Intelligence We address the problem of runtime trajectory anomaly detection, a critical capability for enabling trustworthy LLM agents. Current safety measures predominantly focus on static input/output filtering. However, we argue that ensuring LLM agents reliability requires auditing the intermediate execution process. In this work, we formulate the task of Trajectory Anomaly Detection. The goal is not merely detection, but precise error localization. This capability is essential for enabling efficient rollback-and-retry. To achieve this, we construct TrajBench, a dataset synthesized via a perturb-and-complete strategy to cover diverse procedural anomalies. Using this benchmark, we investigate the capability of models in process supervision. We observe that general-purpose LLMs, even with zero-shot prompting, struggle to identify and localize these anomalies. This reveals that generalized capabilities do not automatically translate to process reliability. To address this, we propose TrajAD, a specialized verifier trained with fine-grained process supervision. Our approach outperforms baselines, demonstrating that specialized supervision is essential for building trustworthy agents. |
| title | TrajAD: Trajectory Anomaly Detection for Trustworthy LLM Agents |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2602.06443 |